DeepSeek V4 Pro
DeepSeek V4 Pro
DeepSeek V4 Pro 0813 is a 1.6T Mixture-of-Experts model with 49B active parameters, built for advanced coding, tool use, and long-horizon agentic work.
About DeepSeek V4 Pro
DeepSeek-V4-Pro-0813 is the official DeepSeek V4 Pro release, replacing the preview with substantially enhanced agentic capabilities and stronger production performance. It is available as a Cloud model in LM Studio Bionic.
At a glance
- 1.6T-parameter Mixture-of-Experts model with 49B active parameters
- 262K-token context in LM Studio Cloud for large codebases and long-running tasks
- Long-horizon agentic performance across coding, terminal, tool-use, and automation benchmarks
- Configurable reasoning with
low,high, andmaxeffort levels
Benchmark results
DeepSeek reports that the official Pro release improves on both the Pro preview and DeepSeek V4 Flash across every benchmark below.
| Benchmark | DeepSeek V4 Pro 0813 | V4 Flash 0731 | V4 Pro Preview | GLM-5.2 | Kimi K3 | Opus-4.8 |
|---|---|---|---|---|---|---|
| HLE (without / with tools) | 42.7 / 60.0 | 37.8 / 51.5 | 37.7 / 48.2 | 40.5 / 54.7 | 43.5 / 56.0 | 49.8 / 57.9 |
| Terminal Bench 2.1 | 87.9 | 82.7 | 72.1 | 81.0 | 88.3 | 85.0 |
| NL2Repo | 61.5 | 54.2 | 38.5 | 48.9 | — | 69.7 |
| Cybergym | 83.3 | 76.7 | 52.7 | — | 80.0 | 78.3 |
| DeepSWE | 62.7 | 54.4 | 12.8 | 46.2 | 67.5 | 58.0 |
| Toolathlon-Verified | 74.1 | 70.3 | 55.9 | 59.9 | 76.5 | 76.2 |
| Agents' Last Exam | 25.7 | 25.2 | 16.5 | 23.8 | 27.6 | 25.7 |
| AutomationBench Public | 31.8 | 25.1 | 12.8 | 12.9 | 30.8 | 27.2 |
| DSBench-FullStack † | 71.1 | 68.7 | 41.8 | 61.8 | 73.7 | 71.6 |
| DSBench-Hard † | 67.2 | 59.6 | 31.1 | 54.5 | 63.0 | 71.7 |
Source: DeepSeek V4 Pro 0813 official model card. For public code-agent tasks, DeepSeek evaluated the model with the minimal mode of DeepSeek Harness at max reasoning effort, temperature = 1.0, and top_p = 0.95. † DSBench-FullStack and DSBench-Hard are internal DeepSeek test sets. Scores and methodology are reproduced as reported by DeepSeek.
Reasoning
Use reasoning_effort to choose between low, high, and max deliberation. DeepSeek recommends temperature = 1.0 and top_p = 0.95 for agentic workloads, and allowing up to 384K output tokens for high and max reasoning where supported.
Use DeepSeek V4 Pro in LM Studio Bionic
Choose DeepSeek V4 Pro from the Cloud model picker in Bionic. LM Studio Cloud inference servers are US-based, and Zero Data Retention is enabled by default.
License and sources
DeepSeek V4 Pro is released under the MIT License.